Self-ratings of health by individuals responding to surveys have shown themselves to be potent predictors of mortality in a growing number of studies. Regardless of the amount or quality of the other health data available in the survey, the individual's global self-rating of health as excellent, good, fair, or poor retains an independent predictive effect on mortality. A key question raised by these studies is, do poor self- assessments increase the risk of mortality only, or do they also increase the risk of morbidity during the remaining years of life? The goal of this project is to assess the impact of self-ratings of health on subsequent health status as measured by functional ability in activities of daily living. The project will start with preliminary findings from the Yale Health and Aging Project, one of the four Established Populations for Epidemiologic Studies of the Elderly (EPESE) sites. These data show that self-ratings of health in 1982, net of baseline functional ability, health, and sociodemographic status, are associated with changes in functional ability over periods of one and two years. The first year of the project will enlarge these findings by analyzing data from each of the additional follow-up years of 1985 through 1988; by doing sex-specific analyses; by taking different approaches to the measurement of functional ability; and by looking for differences between demographic subgroups. The aim will be to see if the initial findings hold for the entire follow-up period, for important subgroups, and for various levels of disability. The second objective with these data will be to better understand the meaning of self- ratings vis-a-vis changes in functioning, by testing various psychosocial measures as explanatory mechanisms. Depression, hope, or optimism about the future, social isolation, religious involvement, and somatic concern, are all possible explanations of the observed effect for which indicators are available in the New Haven EPESE data. The models estimated for the New Haven EPESE data will then be tested with the NHANES I Epidemiologic Follow-up Studies (NHEFS) for 1982-84 and 1987, in which extensive data on functional ability are available. The NHEFS is a nationally representative probability sample, thus the models estimated for an elderly population can be tested with a sample representing all U.S. adults. Excellent psychosocial data are also available in the NHEFS. Together these studies should contribute substantially to our understanding of the meaning of excellent, good, fair and poor ratings of health, and their implications, not just for survival, but for the loss and recovery of functional ability in daily life.